OpenAI has confirmed that the U.S. government will have authority to vet and approve certain users seeking high-level access to its upcoming model, tentatively named GPT-5.6, according to a report from The Washington Post.
Under the arrangement, prospective users seeking elevated access tiers must go through a vetting process run by a U.S. government committee. OpenAI will continue to operate the underlying infrastructure but will not be able to override a government denial for these high-level access requests. The Post frames the move as an unusual formal linkage between a commercial AI company's access controls and federal oversight.
The rationale cited centers on concerns that increasingly capable models could be misused by state-sponsored actors, including in cyberattacks, disinformation, or biological weapons research — concerns intelligence agencies have raised about frontier AI generally.
It's not yet clear from available reporting exactly what criteria the government committee will use, how narrow the 'high-level access' tier is relative to standard usage, or how quickly other AI developers might face similar arrangements. The Washington Post report is the primary source for these details; readers should look there for the government's official framing of the process.